Sneha Mehta, a Ph.D. student in computer science at the Discovery Analytics Center, was in New York City this week to present “Simplify-then-Translate: Automatic Preprocessing for Black-Box Translation” in a talk and poster presentation at the main AAAI Conference on Artificial Intelligence.
The paper represents her work on novel methods to improve machine translation for subtitles while an intern at Netflix for two consecutive summers.
In fact, last summer was a busy one for Mehta, who is advised by Naren Ramakrishnan. In addition to an internship at Neflix headquarters in Los Gatos, California, she was also selected to attend the Deep Learning and Reinforcement Learning Summer School (DLRLSS), in Edmonton, Alberta, Canada.
“I applied to the Summer School because both deep learning and reinforcement learning are very relevant to my work both at DAC and the work I was doing at Netflix,” Mehta said. “Hearing directly from some of the pioneers in the field was a great – and invaluable – experience.”
As an undergraduate, Mehta did a couple of internships where she was introduced to data mining and deep learning. “As a result, I decided I wanted to study these areas in more depth,” Mehta said.
Mehta earned a bachelor’s degree in computer science and a master’s degree in mathematics from BITS Pilani University in India.
“I was attracted to Virginia Tech and DAC to pursue a Ph.D. because of the faculty who are doing cutting edge research in the fields of data science and machine learning,” Mehta said. “What I like best about being a DAC student is the opportunity it provides for interdisciplinary collaborations.”
Last May, Mehta presented a poster of her collaborative work at DAC, “Event Detection using Hierarchical Multi-Aspect Attention,” at The Web Conference.
Recently, Mehta was named a 2020 Twitch Research Fellow, one of five students across the country to receive the $10,000 award to support her academic research and an offer to connect with a team at Twitch for a full-time paid internship at Twitch HQ.
Mehta has presented a successful preliminary of her dissertation on new frameworks for event detection and extraction and is projected to receive her Ph.D. in Fall 2020. After graduation, she would like to have a research role in industry.